P11.11: An automatic algorithm to extract first trimester fetal head and trunk structures from 3D volumes

• DOI: 10.1002/uog.13060
• Publication Date: 2015-10-13T12:40:25Z
• Authors: P. Tsai; Y. Sun; P. Fan; C. Chang
• Keywords: Fetal head; 3D ultrasound; Feature (linguistics)

In the prenatal diagnosis of fetal structures, three-dimensional (3D) ultrasound (US) is useful as it provides a multi-planar view. However, an expert must designate area interest and applicable The measurement structures using 3D US thus usually time-consuming subjective. Thus, we proposed model-based segmentation method to automatically segment head trunk which assist measure parameters for clinical evaluations in first trimester. Fifteen static volumes with gestational ages 11 13 weeks were used study. First, construct statistical shape models both by expert-adjusted mesh shapes from training volume images. Second, some anatomical feature points target image, define local coordinate system initially align model image data. Third, apply Active Shape Model (ASM) mechanism adjust global deformed can mostly fit boundaries. results compared those manually obtained expert. experimental show that overcomes difficulties achieves good consistency between automatic manual measurements. Due treating highly-noise images, fully tool lack. Different previous studies segmentations second or third trimester, focus on analysis fetus precisely measures craniofacial making service. also be applied other fields future.